Go to ground.news/wsm to stay ahead of the algorithm for breaking news stories about finance and market insights. Sign up through my link to get 50% off for unlimited access on the Vantage plan.
@@raybod1775 - Fortunately, they farm out the actual chip fabrication to TSMC. And hopefully they only have modest contractual obligations. Don't need a warehouse full of unsold GPUs, nor a multiyear contract with TSMC. When the bust happens, there will be a lot of surplus GPUs, maybe even sold at the Dollar Tree. Hopefully, Nvidia already got paid for those, before the bust. It will still drop their future sales, especially while that surplus is still in the supply chain.
Basically four companies buy nvidias entire production, and now they are looking to build nuclear plants to supply energy. Similar to diminishing returns of ever greater video game graphics, if the game itself was not engaging/rewarding there is no point.
The biggest problem with these ai services is the same with Uber and Lyft. Nothing is special about any of them so it’s a race to the bottom for costs. So the subscription prices will problem have to be slashed
In what regard? Claude is best for coding and analysis, gpt is fairly prices and reasons quote well, gemini is cheap and offers huge context windows and best multimodal, they almost all serve a purpose and have a niche? if you think these tools have no value youre using them wrong
The existing AI services actually have a huge advantage if they existed prior to the release of ChatGPT. If you had scraped the internet from before then, the quality was much higher. Now it's flooded with bots that run off of ChatGPT. And if you train AI off of AI, the quality drops off.
First of all no one is going to be paying so many different AI subscriptions, people have only so much money to spend. It'll be far worse than streaming services.
@@devilsolution9781 It's not whether they have value, it's whether they're worth the cost in fees as well as time. There's still a fair amount of human oversight and fixing up needed if you want to use the current crop of AIs for real applications. I don't believe that today's fees are actually profitable. Doing inference with trillions of parameters isn't cheap, and neither is the hardware. Never mind that they still need to recoup the training costs along with the R&D budget. Like Uber and Lyft a decade ago, AI vendors are selling at a loss to entice more customers to try out the service.
The AI boom actually shows us that almost every CEO out there is as vulnerable as the SoftBank guy: a charismatic guy present an revolutionary breakthrough that makes no profit, and you dumb billions into it and in the end get nothing out of it.
@@letsburn00I mean Ai & ML do have extraordinary potential, that's why people invest in them, they don't invest solely on current metrics or current projections even. It makes it hard to pinpoint the value of them and they are bubbles, but a prolonged bubble can create a gem. For example many of the current largest companies were unprofitable cash burning startups a couple decades ago. Amazon took 7 years and Tesla 17 to reach profitability
@ The AI boom has placed consumers in competition with huge corporations for the same silicon. This alone wouldn’t be as bad, if Nvidia was not in monopolist position. Nvidia’s prime competitor, AMD has made repeated missteps in GPU development.
@@jwillsher80 AMD has 2 gpu divisions. the reality is that Nvidia used its mindshare and money to push for AI based interpolation as an "requisite" to run games. AMD Datacenters GPUS made them a lot of money.
@@jwillsher80nope monopolistic behavior from nvidia is the main component to AMD's lack of performance. Yes some development mistakes but in general no.
@ Please elaborate of the specific monopolistic behaviors. If you mentioned Intel, I would agree as they have tried to harm AMD in various ways. However, I am unaware of specific acts Nvidia has taken.
There's a huge market rationalisation looming - with big opportunities for short sellers. The rate of technical improvement is already slowing down - all the low hanging fruit has been exploited, and many experts claim that further progress will become exponentially harder. And yet all the non-trivial code it generates is still riddled with bugs, and just yesterday a leading AI screwed up a simple temperature conversion for me by an order of magnitude. There's a limit to how much people will pay for a service that's wrong as much as it's right. Their projected revenues don't seem remotely attainable, unless something changes radically.
When the latest AI model gives me hallucinations for half the specialized questions I ask it you have to wonder what the fuck hundreds of billions of dollars are being poured into this shit for
Bro all the ai companies have yet to find any use for their models, if you say coding well I got bad news almost all programmers unanimously say they spend more time fixing the code generated from LLMs than actual coding. Just watch the primegon basically said the same stuff they’re not replacing programmers any time soon and can’t even follow th context of a technical conversation. Sam Altman realistically comes off as a grifter with his begging for regulations, and declining quality of models since chat gpt 3.5
I'm gonna sell all my Nvidia stock the moment ChatGPT raises their prices to make a profit. No way in hell they will ever make a profit. Even if there is a productive use case for AI, it will be done locally on your device using specialized software, not online with some generic AI model.
We are living through the era of funny money. Fundamentals don't matter. There's enough cash being shifted around that it's like old record companies trying to sign anyone because of FOMO. We should have had a deep market correction by now, but no, the heat keeps getting turned up.
The end of the extremely low interest rate was meant to be the market correction, but nope VC got more money than sense. Blame the LPs and aligning CEO bonus with stock performance.
Oh no. For one, NVidia is not part of the hype - it EARNS on the hype. For two, it has other viable businesses. TESLA is far more overvalued than NVidia - Tesla is niche manufacturer on established market - Nvidia is *THE MARKETS*.
Yeah Nvidia is slightly overpriced because of speculation but they are **winning** here, as they say, the money's in selling shovels during a gold rush.
NVDA is a great business, and their valuation only makes sense if they can maintain the earnings growth they've had over the past couple of years. If their earnings increase, but the rate of increase declines, they will go down in price. I think it is unrealistic for NVDA's earnings rate to do anything but decelerate.
I been in this industry long enough to see the rise and fall of AMD Opteron, the fall of AMD, the dominance and fall of Intel, the return of AMD, and see the four largest datacenter customers operate more than half of worldwide server and all of them begin to phase out x86 for custom-ARM server processors. The same will happen with Nvidia, eventually the Trainium and Inferentia and TPU chips catch up to Nvidia, become 'good enough', and custom AI chips will replace Nvidia, just like how x86 is beginning to be phased out at the hyperscale level today. The hyperscale datacenters are too big for Nvidia to sustain this in the long term.
Like you said at the start, it ultimately comes down to every-day use adoption. Businesses want to cut costs which will result in less employees and unemployed people won’t be able to afford to use it.
It's falling apart for Adobe. The more Adobe pumps their once-functional software full of AI junk, the more companies like mine are seeking software alternatives to Adobe. Adobe's software is useless and slow now
Nvidia is going to keep making significant amounts of money for the foreseeable future, as cloud computing continues to expand and as data centers gradually look to replace their older hardware with newer and somewhat less energy consuming GPUs. But I agree with everything that was said in the video. IMO, NVDA stock is priced at roughly twice what it should be if the markets were rational and not so heavily influenced by gamblers betting on trends and sentiment. I recently sold all of my shares. The price is up a little since then, but I'm happy with the gains that I've locked in.
Trump's proposed tariff plan has caused market volatility, leading to declines in tech and manufacturing stocks. With uncertainty ahead, I'm reallocating assets in my $400K portfolio, aiming to avoid losses while staying positioned for a potential market rebound.
A long-term approach can definitely help with navigating market volatility. Set Clear Goals, Focus on Quality Investments,Stay Patient and Avoid Emotional Reactions, and Work with a Financial Advisor
After selling at a loss during the dip, I was hesitant to reinvest my $600K. My financial advisor created a long-term strategy, focusing on diversification and dollar-cost averaging. In just 18 months, my portfolio grew to $850K. Their guidance has been invaluable in helping me stay steady and think long-term through market changes.
That's amazing! Could you share how I might get in touch with your advisor? I’d love to learn more about their approach and see if they can help me achieve similar results.
*Victoria Louisa Saylor* is the licensed fiduciary I use. Just research the name. You’d find necessary details to work with a correspondence to set up an appointment..
Such a well researched and put together Video about Companies & their Valuation. You're doing an awesome job of analyzing and putting forth arguments of questioning the current state of financial markets. Well Done ! Your content is very valuable ! Hopefully in Billions ! :D
I'm not doubting the overall message of the video, but how long are GPUs realistically used by AI companies until they get replaced by newer, faster ones? Around the 15:36 point the video mentions: "once you buy a GPU, it lasts for many years".
Using chat GPT to get inventory? Or just write some quick sql and save the query. Or build it into an excel sheet or app. Then you can get reliable results consistently
A recent robust experiment set up a community of AI bots with "social messaging". They immediately created delusions, lies, and bitterly opposed factions. My bias is that the Butlerian Jihad was due to "thinking machines" being so counterproductive. I already routinely thank humans for overcoming what the machines are not doing well. Today I had to go through 22 web pages and four texts to get the social insecurity message that there will be 2.5% raise for everybody this year.
@@samsonsoturian6013 but the bots aren't neutral. Ask them about any remotely controversial topic and AI takes the liberal position and downplays or outright ignores the conservative
I think what is very often overlooked is that companies like open AI or anthropic are not going to drive their revenue by selling subscriptions to chat bots. Their goal is to get corporate clients and replace workers where they can charge $1000 or $1500 a month or whatever. That is their game.
@@Joe-yh7eb I think that for a few years comparative advantage will still be something that keeps humans relevant, but there’s gonna come a point when even that won’t be the case… However, I expect that the pressure of competition is going to drive this process and I think that once we see agents that are reasonably capable, which looks like it’ll be 2025, maybe 2026 you’re gonna start to see this
Yup, I've been thinking a lot about AI company valuations the last week NVDA has been tanking in price. Prices are so juiced up, as if current revenues will sustain itself at these elevated levels for an uncertain return. Is more computing power going to meaningfully drive up revenue from what mag 7 has now? Maybe, maybe not, but it is priced as if it certainly will.
@@Cap_management objectively false and forward PE is a garbage indicator since its literally a guess in the dark and gets completely crushed the moment earnings underperform
I asked Grok how much power it takes to generate an image. It pull articles that stated ~0.3kW hours. at $0.25/kWhr that is $0.075. How much money is wasted on stupid AI images that are thrown away? I did 6 images and got bored.
Not only do you need massive amounts of generations to get anything worthwhile, but from personal online experience, overproduction is practically built into the culture in it of itself, with users finding the need to create regularly and post variations or "their unique ideas" literal garbage heap, dude.
LLMs are over-hyped and will probably plateau soon since there isn't all that much unexploited data left in the world; on the contrary, the volume of AI slop on the Internet today is more likely to poison new models. However, neural net models in general are useful for specific applications and have been successfully integrated into many products. They just don't have the same hype as ChatGPT. NVIDIA will still have a market for high-end GPUs after the LLM bubble bursts, albeit not as big. For people contemplating AGI, I think that will require revolutionary changes. Probably a few more cycles at least. Fiddling with LLMs isn't going to get there.
The problem is Regenerative AI is actually just a glorified google search engine. AI doesn't understand the information it's giving you, it doesn't even understand your question - it is pattern matching based on the information you supply. It is a very useful tool to help you in your day-2-day work - but that's all it is - a tool. It can't think, it can't reason, it can't be creative - all it does is serve up information it's found elsewhere.
My experience has been quite different. I find it insanely creative. It’s also not getting data like a search engine. In many ways it’s a kind of super compression of all human data. I think the question of whether it understands is interesting but moot. Utility is all that matters. Even generative ai from several years ago mapped all 200 million folding proteins in a couple of months at deepmind. Previously it took a phd 5 years to do one, so that’s a billion years of phd level work wrapped up in a few months. And that was from old ai that’s incredibly outdated. Many of the covid vaccines were selected using ai, all of nvidia’s new chips were designed with gen ai. The line that “It’s just a… [fill in blank] will be a moving goalpost as capabilities improve.
Finally, an honest ad from Coca-Cola-soulless, AI-generated, and perfectly fitting. Over a century of delivering diabetes on demand, yet now people are upset about the ad? How authentic!
The coke commercial response is what I see whenever something related to AI is shown. You get a real visceral reaction of negativity or positivity. There isn’t a between. It will be a hard sell for awhile especially when job losses mount due to AI.
Unless I missed it, you only covered the revenue from chatgpt corporate and consumer What about the revenue from direct api usage Our company spends 30x monthly on direct api then we do on the corporate chatgpt Unless those costs are also separate, it seems there's a big chunk of revenue not accounted for here
I think in 2yrs, the Ai hype will run out of steam as the Nvidia latest earnings show and google, Microsoft will be left holding the bag and Mr Huang is already cashing in.
I wonder if NVIDIA will be able to go up even higher in 10 years time. 2034, what their stock price will be? 2044, 20 years later, what will it be? if you just look at their monthly chart, it looks very parabolic and seems like it has hit the wall. any parabolic thing will never end good. never seen a chart that after being parabolic the price stays up high forever.
The tech behind GPUs is not new. Its just a small fast CRAY supercomputer. It should be commodity hardware like the old beige PCs. The reason NVIDA popped is that their GPUs were the easiest to program so a bunch of ML/AI libraries only worked on NVIDA. That is changing. Even if the new ones take a few years to get up to speed, the performance/cost ratio will kill NVIDA's party.
It’s crazy now that I’m deep into local LLMs… all this is free 😂 in fact it’s more powerful than cloud once you kit it out. I’m glad they got cash monies to keep pushing state forward though. These local models would not exist without GPT starting a fire 🔥
I've been saying this since the beginning, jobs being replaced depends on whether these services are faster and cheaper than how we were doing things before. People talk like they're fast and cost nothing. They're just getting more expensive to make train and run. And thier solution rather than make it more efficient is to try to revolutionize nuclear fusion. If they succeed in that, it's going to change a bunch of things that will have a much bigger impact on society than chatbots and making commercials cheaper to produce. It's a bubble and it's going to burst.
@@Poochie1 never selling (edit: not trying to be cute, but i have had to ignore people and comments like you by the thousands over 5 years on my road to literal multi millions, with peace and love ;) 💚🖤📈🇺🇸
In the 90s they thought we would need cables two feet in diameter to transmit HD video to every home. Then they invented video compression. AI will too become more efficient, so the costs won’t be the same in the future as they are now.
ok good video. i’m forced to mention that you left out a major puzzle piece….sovereign ai (government and military) which has not even been touched or factored into sales yet. that’s all. 💚🖤
@ it is immediately clear from this statement that you don’t even know enough on this subject to even try to pretend you do. do you know what i mean by this?
AI stocks will dominate 2025. I think the next big thing will be A.I. For enduring growth akin to META, it's vital to avoid impulsive decisions driven by short-term fluctuations.
Investing Is more than reading quarterly reports. Learnt this from reading Peter Lynch's book. I believe there are people who do this for a living, and I just delegate the task to these professionals. That's how I make money from the market to be honest.
Sophie Lynn Carrabus is the advisr I use and I'm just putting this out here because you asked. You can Just search the name. You’d find necessary details to work with to set up an appointment.
I appreciate it. After searching her name online and reviewing her credentials, I'm quite impressed. I've contacted her as I could use all the help I can get. A call has been scheduled.
I dont see the demand for GPU going down , they are in a race to improve AI and it's gonna cost a lot of GPUs. Im just surprised NVIDIA has so little competition
GPUs aren't optimized for AI, though. There's always a risk that someone will come up with an entirely new architecture that's more efficient in terms of computation and power consumption, making GPUs obsolete (for AI and machine learning, that is). For example, Groq and their chips.
you might also want to extrapolate the amount of electricity that would be required to power all this theoretical AI, I don't imagine those sums would add up either
Co-pilot? Wots that then!? 🤪 But seriously, even though Microsoft has forced this down all our throats, I simply REFUSE to use it! I'm not off-loading my critical thinking abilities to some algorithm! Like the rest of the sacrificial sheep!
FYI: if you pay $157b for OpenAi today, you obviously want a future return from that. The sharemarket returns about 10% pa in the long run, and you'd expect that from large blue chip companies, but that is by no means amazing. So, to achieve that 10% pa over the next 7 yrs the value would need to be $314b, and at that same multiple of 20x, $16b in NPAT. Note: from that point, even more growth would be needed to justify that 20x multiples as companies with stagnant earnings trade at lower multiples than 20. So, today the valuation looks absurd, but when you extrapolate even a modest return into the future it becomes comical.
Funniest part of this is that the current "AI" boom has genuinely been having some revolutionary effects, just not in any of the dystopian ways the AI companies are pushing.
It seems excessive to demand a 20 P/E for the two companies dominating the segment with a massive advantage that could lead to a consolidated position. Amazon who dominates online retailing trades at 44 PE currently
I don't think using B2C revenue to evaluate these companies is a good way. OpenAI and Anthropic should eventually get their most revenue from enterprise users via API. People are building multi agent system using these models. And these base models are literally the brains of the agents.
Whether or not this analogy is correct, yesterday I watched a program about scientists concerned about a population explosion back in the 60s or so. Thus, countries such as China came up with their one child policy. How wrong they were. Were there policies and theories back then that did become true ? Exponentional a I growth or not.
There is little doubt AI will be the future. Just like internet bubble, it is just overhyped too early by about 10-15 years. Let's just be honest, it's pretty obvious many jobs will become be increasingly autonomous, just like autonomous driving cars and planes are already a reality just not yet in a large scale, like the internet in the 90s. When all the cars on the street drives themselves, all the Amazon workers and car manufacturers are bunch of robots, think of the computing power it would require. More autonomy, less human labor, is as certain as the aging and shrinking labor force in most countries. Yes, AI is overhyped relative to what it can do, but let's not be so blind as to recognize which direction the human civilization is going.
Internet was more like electricity, AI is mostly just slop. Don't compare everything to previous inventions, we've been stagnating for the past 15 years or so, just look at CPUs how little they progress.
Go to ground.news/wsm to stay ahead of the algorithm for breaking news stories about finance and market insights. Sign up through my link to get 50% off for unlimited access on the Vantage plan.
it's the new bitcoin
How is this a day old? Was the video private for a whole day?
@@wallstreetmillennial Groundnews is flooding every single channel I watch. It’s getting annoying.
At this point, an AI car wash company would reach a 1 billion market cap before it ever washes a single car =/
But it’s gonna have reoccurring revenue bro
try reading an income statement
Pre-revenue
Stonk guys would ask you "what is that AI car wash company????? can you give me the ticker plz? i have 1000 dollars to FOMO on it"
Hey that's a good idea, let's do that!
A gold rush where the only people making money are the ones selling picks and shovels.
Like Nvidia
@@raybod1775 - Fortunately, they farm out the actual chip fabrication to TSMC. And hopefully they only have modest contractual obligations. Don't need a warehouse full of unsold GPUs, nor a multiyear contract with TSMC. When the bust happens, there will be a lot of surplus GPUs, maybe even sold at the Dollar Tree. Hopefully, Nvidia already got paid for those, before the bust. It will still drop their future sales, especially while that surplus is still in the supply chain.
Palantir might be the gold
"those are just dumb shiny rocks. Why would anybody work so hard to get them out of the ground"
-you
@@spiritrider963 those guys lost money too
You mean they weren't absurd before?
data centres quite literally power the world
Basically four companies buy nvidias entire production, and now they are looking to build nuclear plants to supply energy. Similar to diminishing returns of ever greater video game graphics, if the game itself was not engaging/rewarding there is no point.
the longer the bull run, the harder the fall.
I work in the “Ai” space creating compute servers, and yeah I’m nervous for when all goes poof
I bet you're making a killing tho
Dot com 2.0
@ Eh… I make decent money but I’m more concerned about job security than anything. I’m on a work visa so getting laid off is terrifying
Marry a citizen @@tansieten875
Can I join @@tansieten875
The biggest problem with these ai services is the same with Uber and Lyft. Nothing is special about any of them so it’s a race to the bottom for costs. So the subscription prices will problem have to be slashed
In what regard? Claude is best for coding and analysis, gpt is fairly prices and reasons quote well, gemini is cheap and offers huge context windows and best multimodal, they almost all serve a purpose and have a niche? if you think these tools have no value youre using them wrong
The existing AI services actually have a huge advantage if they existed prior to the release of ChatGPT. If you had scraped the internet from before then, the quality was much higher. Now it's flooded with bots that run off of ChatGPT. And if you train AI off of AI, the quality drops off.
First of all no one is going to be paying so many different AI subscriptions, people have only so much money to spend. It'll be far worse than streaming services.
@@devilsolution9781 It's not whether they have value, it's whether they're worth the cost in fees as well as time. There's still a fair amount of human oversight and fixing up needed if you want to use the current crop of AIs for real applications.
I don't believe that today's fees are actually profitable. Doing inference with trillions of parameters isn't cheap, and neither is the hardware. Never mind that they still need to recoup the training costs along with the R&D budget. Like Uber and Lyft a decade ago, AI vendors are selling at a loss to entice more customers to try out the service.
Perplexity or abacus use API from all major AI model. Streaming is different than AI subscription.@@thomasfalcon6350
The AI boom actually shows us that almost every CEO out there is as vulnerable as the SoftBank guy: a charismatic guy present an revolutionary breakthrough that makes no profit, and you dumb billions into it and in the end get nothing out of it.
Yea man all those stupid ceos. They should hire you insted.
You know it's a bubble when the existence of an industry drains others of capital
Do you mean you donˋt want my E-Book about MLM?
I work for an engineering based startup and seeing the groups getting money thrown at them while actually inventions take lots of effort is nuts
@@letsburn00I mean Ai & ML do have extraordinary potential, that's why people invest in them, they don't invest solely on current metrics or current projections even. It makes it hard to pinpoint the value of them and they are bubbles, but a prolonged bubble can create a gem.
For example many of the current largest companies were unprofitable cash burning startups a couple decades ago. Amazon took 7 years and Tesla 17 to reach profitability
@@letsburn00they would probably do better if they hired people that can write a single proper sentence
@@WhatSmellsLikeToast My apologies that I'm not writing Proust in my RUclips comments.
This is also negatively impacting gaming GPUs. It is like the crypto absurdity 2.0.
How?
@ The AI boom has placed consumers in competition with huge corporations for the same silicon. This alone wouldn’t be as bad, if Nvidia was not in monopolist position. Nvidia’s prime competitor, AMD has made repeated missteps in GPU development.
@@jwillsher80 AMD has 2 gpu divisions. the reality is that Nvidia used its mindshare and money to push for AI based interpolation as an "requisite" to run games. AMD Datacenters GPUS made them a lot of money.
@@jwillsher80nope monopolistic behavior from nvidia is the main component to AMD's lack of performance. Yes some development mistakes but in general no.
@ Please elaborate of the specific monopolistic behaviors. If you mentioned Intel, I would agree as they have tried to harm AMD in various ways. However, I am unaware of specific acts Nvidia has taken.
Like all bubbles eventually it will burst, the trick is to guess how big it will get just before that burst
The market has been irrational on Tesla' stock price for years now, I'm sure reheated AI promises will be around for a long time.
There is no bubble. Those companies are undervalued
There's a huge market rationalisation looming - with big opportunities for short sellers. The rate of technical improvement is already slowing down - all the low hanging fruit has been exploited, and many experts claim that further progress will become exponentially harder. And yet all the non-trivial code it generates is still riddled with bugs, and just yesterday a leading AI screwed up a simple temperature conversion for me by an order of magnitude. There's a limit to how much people will pay for a service that's wrong as much as it's right. Their projected revenues don't seem remotely attainable, unless something changes radically.
When the latest AI model gives me hallucinations for half the specialized questions I ask it you have to wonder what the fuck hundreds of billions of dollars are being poured into this shit for
It's a gold rush and Nvidia is making shovels.
Bro all the ai companies have yet to find any use for their models, if you say coding well I got bad news almost all programmers unanimously say they spend more time fixing the code generated from LLMs than actual coding. Just watch the primegon basically said the same stuff they’re not replacing programmers any time soon and can’t even follow th context of a technical conversation. Sam Altman realistically comes off as a grifter with his begging for regulations, and declining quality of models since chat gpt 3.5
Sam Altman needs to be in jail really.
Not to mention, that those who can claim CODING takes most of their job, can count themselves lucky.
Uhhh I use ChatGPT every day.
AI is a good tool for coding, it can help with ideation and debugging. It does not replace actually software engineering skills
The code GPT spits out can be pretty bad. I like it as a learning resource, but the code has consistently been pretty bad in my experience.
I'm gonna sell all my Nvidia stock the moment ChatGPT raises their prices to make a profit. No way in hell they will ever make a profit. Even if there is a productive use case for AI, it will be done locally on your device using specialized software, not online with some generic AI model.
Hardware is the problem. No one is going to run 80k gpus locally. Far more efficient to run it all through huge data centers.
let's go!
My favourite economic doom & gloom channel back at it
Nah that would be How Money Works. WSM has a very moderate way of looking at it.
incorrect
they make a good bear case to be fair. What is your bull case?
Probably thinking of Moon
@@The_Orgazoid I don't mind the channel, I like his analysis; he is just my main bear analysis guy
Altman is like Neumann from WeWork - talking up his book on the way down. Typical scammer.
We are living through the era of funny money. Fundamentals don't matter. There's enough cash being shifted around that it's like old record companies trying to sign anyone because of FOMO. We should have had a deep market correction by now, but no, the heat keeps getting turned up.
The end of the extremely low interest rate was meant to be the market correction, but nope VC got more money than sense. Blame the LPs and aligning CEO bonus with stock performance.
Great analysis. Your subscribers will cherish this video when they look back in 9 months time.
It's such a poorly done video, I don't know what you're applauding him for.
@@JudeMarchisio explain or are you just an AI moon boy huffing copium?
Oh no. For one, NVidia is not part of the hype - it EARNS on the hype. For two, it has other viable businesses. TESLA is far more overvalued than NVidia - Tesla is niche manufacturer on established market - Nvidia is *THE MARKETS*.
Yeah Nvidia is slightly overpriced because of speculation but they are **winning** here, as they say, the money's in selling shovels during a gold rush.
NVDA is a great business, and their valuation only makes sense if they can maintain the earnings growth they've had over the past couple of years. If their earnings increase, but the rate of increase declines, they will go down in price. I think it is unrealistic for NVDA's earnings rate to do anything but decelerate.
The bursting of the AI bubble will make the late 90s tech bubble look like Sunday brunch.
Company just introduced AI Robot girlfriend that's indistinguishable from real life. Valuation goes parabolic.
I been in this industry long enough to see the rise and fall of AMD Opteron, the fall of AMD, the dominance and fall of Intel, the return of AMD, and see the four largest datacenter customers operate more than half of worldwide server and all of them begin to phase out x86 for custom-ARM server processors. The same will happen with Nvidia, eventually the Trainium and Inferentia and TPU chips catch up to Nvidia, become 'good enough', and custom AI chips will replace Nvidia, just like how x86 is beginning to be phased out at the hyperscale level today. The hyperscale datacenters are too big for Nvidia to sustain this in the long term.
How many years from now… 2 or 20?
A more apt comparison is crypto mining, ASICs have replaced GPUs for mining, but crypto mining is basically useless.
Like you said at the start, it ultimately comes down to every-day use adoption. Businesses want to cut costs which will result in less employees and unemployed people won’t be able to afford to use it.
People will desperately pay $80 a month for an AI that improves their job performance enough so they don't get fired.
@@skierpage if job need 100 normally but Ai efficiency makes it 10. now 90 unemployed.
@skierpage that assumes AI is threating them to begin with.
Maybe people will go back to constructing housing.
Adobe has 33 million paying subscribers for their robust tools. Good luck achieving that number without making the product far more compelling.
In ten years Adobe will be redundant as AI will be able to do all of that stuff
@@method341 Adobe will have deep AI interfacing soon, they won't be sitting duck
It's falling apart for Adobe. The more Adobe pumps their once-functional software full of AI junk, the more companies like mine are seeking software alternatives to Adobe. Adobe's software is useless and slow now
Adobe targets a very niche market of creative professionals while chatgpt is more of a mass market product
Nvidia is going to keep making significant amounts of money for the foreseeable future, as cloud computing continues to expand and as data centers gradually look to replace their older hardware with newer and somewhat less energy consuming GPUs. But I agree with everything that was said in the video. IMO, NVDA stock is priced at roughly twice what it should be if the markets were rational and not so heavily influenced by gamblers betting on trends and sentiment. I recently sold all of my shares. The price is up a little since then, but I'm happy with the gains that I've locked in.
Good call, congrats on the profit you've taken!
Trump's proposed tariff plan has caused market volatility, leading to declines in tech and manufacturing stocks. With uncertainty ahead, I'm reallocating assets in my $400K portfolio, aiming to avoid losses while staying positioned for a potential market rebound.
A long-term approach can definitely help with navigating market volatility. Set Clear Goals, Focus on Quality Investments,Stay Patient and Avoid Emotional Reactions, and Work with a Financial Advisor
After selling at a loss during the dip, I was hesitant to reinvest my $600K. My financial advisor created a long-term strategy, focusing on diversification and dollar-cost averaging. In just 18 months, my portfolio grew to $850K. Their guidance has been invaluable in helping me stay steady and think long-term through market changes.
That's amazing! Could you share how I might get in touch with your advisor? I’d love to learn more about their approach and see if they can help me achieve similar results.
*Victoria Louisa Saylor* is the licensed fiduciary I use. Just research the name. You’d find necessary details to work with a correspondence to set up an appointment..
I’ve just looked up her full name on my browser and found her webpage, very much appreciate this
This channel is so informative, among the best on RUclips. Thank you!
Such a well researched and put together Video about Companies & their Valuation. You're doing an awesome job of analyzing and putting forth arguments of questioning the current state of financial markets. Well Done ! Your content is very valuable ! Hopefully in Billions ! :D
I'm not doubting the overall message of the video, but how long are GPUs realistically used by AI companies until they get replaced by newer, faster ones?
Around the 15:36 point the video mentions: "once you buy a GPU, it lasts for many years".
Using chat GPT to get inventory? Or just write some quick sql and save the query. Or build it into an excel sheet or app. Then you can get reliable results consistently
Well thank you for picking ground news as the sponsor
How would AI knows if its fake news or facts?
A recent robust experiment set up a community of AI bots with "social messaging".
They immediately created delusions, lies, and bitterly opposed factions.
My bias is that the Butlerian Jihad was due to "thinking machines" being so counterproductive.
I already routinely thank humans for overcoming what the machines are not doing well.
Today I had to go through 22 web pages and four texts to get the social insecurity message that there will be 2.5% raise for everybody this year.
The bot generally avoids taking a side when it finds conflicting info, and uses terms like "complex issue" because humans like neutral bots
@@samsonsoturian6013 but the bots aren't neutral. Ask them about any remotely controversial topic and AI takes the liberal position and downplays or outright ignores the conservative
@Noah_E liar.
They dont.
I think what is very often overlooked is that companies like open AI or anthropic are not going to drive their revenue by selling subscriptions to chat bots. Their goal is to get corporate clients and replace workers where they can charge $1000 or $1500 a month or whatever. That is their game.
Only a greedy idiot thinks it will work though. Only out of touch elitists who aren't actually investigating anything think this will ever be possible
Then they will realize too late that AI cannot entirely replace a human being.
@@Joe-yh7eb I think that for a few years comparative advantage will still be something that keeps humans relevant, but there’s gonna come a point when even that won’t be the case… However, I expect that the pressure of competition is going to drive this process and I think that once we see agents that are reasonably capable, which looks like it’ll be 2025, maybe 2026 you’re gonna start to see this
Yup, I've been thinking a lot about AI company valuations the last week NVDA has been tanking in price. Prices are so juiced up, as if current revenues will sustain itself at these elevated levels for an uncertain return. Is more computing power going to meaningfully drive up revenue from what mag 7 has now? Maybe, maybe not, but it is priced as if it certainly will.
Price is in line with past 20 years. It's now actually cheap looking at forward PE.
Stocks get overbought, then oversold
@@Cap_management objectively false and forward PE is a garbage indicator since its literally a guess in the dark and gets completely crushed the moment earnings underperform
I asked Grok how much power it takes to generate an image. It pull articles that stated ~0.3kW hours. at $0.25/kWhr that is $0.075. How much money is wasted on stupid AI images that are thrown away? I did 6 images and got bored.
I read somewhere that one chat got response is 100x more expensive than a google search. This is a boondoggle
I read somewhere that Earth is flat and that prices of new technology are flat forever.
Not only do you need massive amounts of generations to get anything worthwhile, but from personal online experience, overproduction is practically built into the culture in it of itself, with users finding the need to create regularly and post variations or "their unique ideas" literal garbage heap, dude.
@@JJ-zr6fu that means nothing if you don't know what a Google search costs
I read somewhere that Earth has finite resources and that technology without energy is sculpture.
Man your voice have become a go to for good small break downs on subjects. Love them. Keep it up man
Thanks for articulating my gut sense on AI.
Bubbles are always amusing to watch; they're much more amusing to watch when they explode.
LLMs are over-hyped and will probably plateau soon since there isn't all that much unexploited data left in the world; on the contrary, the volume of AI slop on the Internet today is more likely to poison new models.
However, neural net models in general are useful for specific applications and have been successfully integrated into many products. They just don't have the same hype as ChatGPT. NVIDIA will still have a market for high-end GPUs after the LLM bubble bursts, albeit not as big.
For people contemplating AGI, I think that will require revolutionary changes. Probably a few more cycles at least. Fiddling with LLMs isn't going to get there.
The problem is Regenerative AI is actually just a glorified google search engine. AI doesn't understand the information it's giving you, it doesn't even understand your question - it is pattern matching based on the information you supply. It is a very useful tool to help you in your day-2-day work - but that's all it is - a tool. It can't think, it can't reason, it can't be creative - all it does is serve up information it's found elsewhere.
My experience has been quite different. I find it insanely creative. It’s also not getting data like a search engine. In many ways it’s a kind of super compression of all human data. I think the question of whether it understands is interesting but moot. Utility is all that matters. Even generative ai from several years ago mapped all 200 million folding proteins in a couple of months at deepmind. Previously it took a phd 5 years to do one, so that’s a billion years of phd level work wrapped up in a few months. And that was from old ai that’s incredibly outdated. Many of the covid vaccines were selected using ai, all of nvidia’s new chips were designed with gen ai. The line that “It’s just a… [fill in blank] will be a moving goalpost as capabilities improve.
My mouse clicks go to closing out annoying co-pilot at work. Great productivity killer
Finally, an honest ad from Coca-Cola-soulless, AI-generated, and perfectly fitting. Over a century of delivering diabetes on demand, yet now people are upset about the ad? How authentic!
02:41 ad skip
Nvidia is gonna make 210 billion revenue in 2025. Its margin will be 75% and that is 157 billion. Take that number
You know its the bottom when everyone says its the top
The coke commercial response is what I see whenever something related to AI is shown. You get a real visceral reaction of negativity or positivity. There isn’t a between. It will be a hard sell for awhile especially when job losses mount due to AI.
Unless I missed it, you only covered the revenue from chatgpt corporate and consumer
What about the revenue from direct api usage
Our company spends 30x monthly on direct api then we do on the corporate chatgpt
Unless those costs are also separate, it seems there's a big chunk of revenue not accounted for here
the market can stay irrational longer than you can stay solvent
being early is the same as being wrong :(
Nvda has a growing software business both ai and non ai with considerable lower margins. Also, gaming buying back their shares will help.
Why do you assume 20x P/E though?
I think in 2yrs, the Ai hype will run out of steam as the Nvidia latest earnings show and google, Microsoft will be left holding the bag and Mr Huang is already cashing in.
AI is not hype. There are limitations and those that can are working on ways to get AI past these limitations.
@ meant hype as in the stock market where money is being thrown at and OpenAI will regret not going public sooner.
I wonder if NVIDIA will be able to go up even higher in 10 years time. 2034, what their stock price will be? 2044, 20 years later, what will it be? if you just look at their monthly chart, it looks very parabolic and seems like it has hit the wall. any parabolic thing will never end good. never seen a chart that after being parabolic the price stays up high forever.
Subscribed
The Bazinga Bust can't come soon enough.
Does openai also generate revenue from their API service? Seems like this was left out of the video.
Should I sell all my nvidia stock
Are there any indicators for when the big 3 cloud providers will finish updating their infrastructure?
The tech behind GPUs is not new. Its just a small fast CRAY supercomputer. It should be commodity hardware like the old beige PCs. The reason NVIDA popped is that their GPUs were the easiest to program so a bunch of ML/AI libraries only worked on NVIDA. That is changing. Even if the new ones take a few years to get up to speed, the performance/cost ratio will kill NVIDA's party.
This is technology frenzy driven not necessarily by scientists and engineers, but Wall Street gurus and financial bean counters.
It’s crazy now that I’m deep into local LLMs… all this is free 😂 in fact it’s more powerful than cloud once you kit it out. I’m glad they got cash monies to keep pushing state forward though. These local models would not exist without GPT starting a fire 🔥
Nvidia is making actual money from producing actual working products.
I've been saying this since the beginning, jobs being replaced depends on whether these services are faster and cheaper than how we were doing things before. People talk like they're fast and cost nothing. They're just getting more expensive to make train and run. And thier solution rather than make it more efficient is to try to revolutionize nuclear fusion. If they succeed in that, it's going to change a bunch of things that will have a much bigger impact on society than chatbots and making commercials cheaper to produce. It's a bubble and it's going to burst.
The higher the climb, the lower the drop.
i am a long time nvidia stan and concentrated shareholder and ai enthusiast so i’ll sit out any editorial commentary ;) 💚🖤📈🇺🇸
happy thanksgiving 🦃🍁
take your money and run.
@@Poochie1 never selling (edit: not trying to be cute, but i have had to ignore people and comments like you by the thousands over 5 years on my road to literal multi millions, with peace and love ;) 💚🖤📈🇺🇸
@@matt.stevick 😂
@@Poochie1 the main question is - where to? this bubble inflates everything else along with it
@@matt.stevick never forget the scam makes money when it goes up, but also when it goes down
In the 90s they thought we would need cables two feet in diameter to transmit HD video to every home. Then they invented video compression. AI will too become more efficient, so the costs won’t be the same in the future as they are now.
The 2025 strategy: primary asset holders induce hysteria in markets to siphon out remaining value before boarding their spaceships.
i just watched a video of Kamala Harris drinking and driving after the election results. I love this tech
and Biden playing the bongos while Trump danced.
Hallucination is the biggest issue. An AI that can accurately cite would be great
ok good video. i’m forced to mention that you left out a major puzzle piece….sovereign ai (government and military) which has not even been touched or factored into sales yet. that’s all. 💚🖤
also not forget civilian applications for vehicles like autonomous driving.
Because it doesn't exist.
@@samsonsoturian6013 what doesn’t exist?
@matt.stevick there's no military demand for product, and what AI exists is being done in-house
@ it is immediately clear from this statement that you don’t even know enough on this subject to even try to pretend you do. do you know what i mean by this?
You forgot about the developer platform, they get revenue from transactional pricing plan too.
The problem is the price. To switch on one of our programs for 30 days costs $86000. That is just to run it
I thought AI stood for All Indians.
Actually Indians
Allen Iverson
A.I bubble will burst soon.
AI stocks will dominate 2025. I think the next big thing will be A.I. For enduring growth akin to META, it's vital to avoid impulsive decisions driven by short-term fluctuations.
Prioritize patience and a long-term perspective consider financial advisory for informed buying and selling decisions.
Investing Is more than reading quarterly reports. Learnt this from reading Peter Lynch's book. I believe there are people who do this for a living, and I just delegate the task to these professionals. That's how I make money from the market to be honest.
@@PatrickLloyd- I've been getting suggestions to use one, but where and how to find one has been challenging, Can i reach out to the one you use?
Sophie Lynn Carrabus is the advisr I use and I'm just putting this out here because you asked. You can Just search the name. You’d find necessary details to work with to set up an appointment.
I appreciate it. After searching her name online and reviewing her credentials, I'm quite impressed. I've contacted her as I could use all the help I can get. A call has been scheduled.
I dont see the demand for GPU going down , they are in a race to improve AI and it's gonna cost a lot of GPUs. Im just surprised NVIDIA has so little competition
GPUs aren't optimized for AI, though. There's always a risk that someone will come up with an entirely new architecture that's more efficient in terms of computation and power consumption, making GPUs obsolete (for AI and machine learning, that is). For example, Groq and their chips.
In a gold rush, be the one selling shovels
A lot of the shovel vendors also went bust during the gold rush.
Make no mistake AI is a thing. LLMs are not. Think the internet in 1998.
you might also want to extrapolate the amount of electricity that would be required to power all this theoretical AI, I don't imagine those sums would add up either
That’s why they are going nuclear
@@squirrel9760agreed
I think you bring up an excellent question. I doubt that new nuclear power plants will be up and running soon enough.
How long has computer boom and internet boom gone on? Why would AI be any different?
The internet grew from 1996-2004 plenty of people lost their money permanently.
Three years since chatgpt? I count 2
We have copilot pro on our work laptops it’s been there for months but most days I forget it’s there.
Co-pilot? Wots that then!? 🤪 But seriously, even though Microsoft has forced this down all our throats, I simply REFUSE to use it! I'm not off-loading my critical thinking abilities to some algorithm! Like the rest of the sacrificial sheep!
FYI: if you pay $157b for OpenAi today, you obviously want a future return from that.
The sharemarket returns about 10% pa in the long run, and you'd expect that from large blue chip companies, but that is by no means amazing.
So, to achieve that 10% pa over the next 7 yrs the value would need to be $314b, and at that same multiple of 20x, $16b in NPAT.
Note: from that point, even more growth would be needed to justify that 20x multiples as companies with stagnant earnings trade at lower multiples than 20.
So, today the valuation looks absurd, but when you extrapolate even a modest return into the future it becomes comical.
Sounds like a bubble, like a lot of FOMO. How long before it pans out, if it ever will? And if it doesn't, how long before the bubble bursts?
Thou shalt not suffer the Abominable Intelligence to exist
Funniest part of this is that the current "AI" boom has genuinely been having some revolutionary effects, just not in any of the dystopian ways the AI companies are pushing.
AI is just a glorified search engine. It just returns search results couched in normal sentences.
No, it isn't. Search engines don't make up search results to sources that don't exist.
@@arotobo If you need AI to help you with those tasks then you are in the wrong job or just not educated enough.
Puts on all these companies.
It seems excessive to demand a 20 P/E for the two companies dominating the segment with a massive advantage that could lead to a consolidated position. Amazon who dominates online retailing trades at 44 PE currently
I wonder if any of the current AI darlings will be the ones to ultimately make it. Definitely a lot will not.
Bubble.
I really like your videos
I pay for the upgraded service to help me write papers for college if I wasn’t in school I wouldn’t pay
I don't think using B2C revenue to evaluate these companies is a good way. OpenAI and Anthropic should eventually get their most revenue from enterprise users via API. People are building multi agent system using these models. And these base models are literally the brains of the agents.
Whether or not this analogy is correct, yesterday I watched a program about scientists concerned about a population explosion back in the 60s or so. Thus, countries such as China came up with their one child policy. How wrong they were. Were there policies and theories back then that did become true ? Exponentional a I growth or not.
Moral of the story: Stonks only go up
There is little doubt AI will be the future. Just like internet bubble, it is just overhyped too early by about 10-15 years. Let's just be honest, it's pretty obvious many jobs will become be increasingly autonomous, just like autonomous driving cars and planes are already a reality just not yet in a large scale, like the internet in the 90s. When all the cars on the street drives themselves, all the Amazon workers and car manufacturers are bunch of robots, think of the computing power it would require. More autonomy, less human labor, is as certain as the aging and shrinking labor force in most countries. Yes, AI is overhyped relative to what it can do, but let's not be so blind as to recognize which direction the human civilization is going.
I wouldn't be so sure to say 10-15 years. We went from a dumb high-schooler to winning math olympiads in 2 years with language models
Internet was more like electricity, AI is mostly just slop. Don't compare everything to previous inventions, we've been stagnating for the past 15 years or so, just look at CPUs how little they progress.
@@thomasfalcon6350AI will be way bigger than the internet.
Some people fear the impacts of AI taking over. I'm more worried about the economic crash when one of the largest speculative bubbles finally pops.
I play poker with the Asian guy in the jacket. He told me he works at whole foods market.
"In order for the web to be profitable, each member of society must pay 300eur in broadband fees, therefore it will crash"
Famous last words!
AI is very good for search engines. I used them all the time. I am the end user, but I don’t pay a dime for it.